The present invention relates to a database management system for searching a database (DB), a database management method and a database management program.
Relational databases can store a large amount of various data, as that in data warehouses. Each piece of data included in a relational database is handled on the basis of a concept referred to as relationship model. Each piece of data is provided in a column and a record (row), and thus placed in each table and organized. Data can be reorganized on the basis of the respective columns and records, so that the operation of sampling data can be made easy. The system for operating/managing a relational database is referred to as relational database management system (RDBMS).
In the case where data is sampled from a relational database, it is necessary to specify the location where the data is stored and the specification of the data by referring to the program specification of the database. In the case where only the column of a specific data type is referred to in a table recorded in a database, for example, SQL sentences designated by the column are programmed using the data type of the column described in the specification, and thus, data is searched. In this case, it is necessary to specify the data the user desires to refer to by directly designating the name of the column or using the view function of the database at the time of programming.
In the case where the name of the column is directly designated, it is necessary to designate a plurality of columns using SQL sentences, which is troublesome. In addition, in the case where a column is added, for example, it is necessary to modify the SQL sentences, and therefore, flexible measures cannot be taken.
Likewise, in the case where the data the user desires to refer to is specified using the view function of the database, it is necessary to prepare a view for each unit of a plurality of data types, and therefore, the preparation is an energy-consuming task and preparation of a search formula takes time and effort, and thus, data cannot be searched smoothly.
In order to make searching of such a database more convenient, a technology using meta-data, which is conceptual information, is discussed in Japanese Laid-Open Patent Publication No. 10-91503. According to the technology in the above described publication, basic management information is registered in an information source management dictionary in advance, in order to collect information on the database system connected to the network. The information source management dictionary manages meta-information on the entirety of the database system. An information source management control system collects meta-information from meta-information on individual database systems and individual information source tools in accordance with timing designated by basic information for collecting information. The information source management control system reflects the collected meta-information in the information source management dictionary.
Many pieces of data can be stored in a data warehouse, in various formats. Therefore, it is difficult to search a plurality of databases having different structures and data names collectively and in a crossover manner in a data warehouse. In the case where the cause and effect relationship of data is examined, for example through data mining, specific types of data cannot be sampled irrespectively of the column name, and therefore, the efficiency of operation lowers.
In addition, in the case where data stored in each table of a relational database is classified, it is necessary to prepare different tables or manage the tables by providing attribute columns as the columns of the tables.
In the technology of the above described publication, the amount of meta-information for classification increases, and it is necessary to prepare complex SQL sentences in the case of classification in a plurality of dimensions. Furthermore, it is difficult to flexibly change the attributes for classification.